Many causal factors to marine traffic accidents (MTA) influence each other and have associated effects. It is necessary to quantify the correlation path mode of these factors to improve accident prevention measures and their effects. In the application of human factors to the accident mechanisms, the complex structural chains on causes to MTA systems were analyzed combining the Human Failure Analysis and Classification System (HFACS) with theoretical Structural Equation Modeling (SEM). First, the accident causation model was established as a human error analysis classification in sight of MTA, and the constituent elements of the causes of accident was conducted. Second, a hypothetical model of Human factors classification was proposed applying the practice of the structural model. Third, with the data resource from ship accident cases, this hypothetical model was discussed and simulated, and as a result the relationship path dependency mode between the latent independent variable of the accident was quantitatively analyzed based on the observed dependent variable of human behaviors. Application examples show that relationships in HFACS are verified and in line with the path developing mode, and resource management factors have a pronounced influence and a strong relevance to the causal chain of the accidents. Appropriate algorithms for the theoretical model can be used to numerically understand the safety performance of marine traffic systems under different parameters through mathematical analysis. Hierarchical assumptions in the HFACS model are quantitatively verified.